Live sports broadcasts are among the most challenging environments for real-time translation and captioning. The roar of the crowd, on-field sounds, music, and multiple commentators all compete within the audio mix, making it extremely difficult for even the most sophisticated speech recognition models to accurately extract dialogue for transcription and translation.

To address this, AI-Media integrated AudioShake’s real-time Dialogue, Music, and Effects (DME) separation technology directly into its LEXI Voice workflow. AudioShake’s SDK allows AI-Media to isolate the announcer’s spoken content with remarkable precision while preserving the immersive background audio that defines the live sports experience.

Once the announcer dialogue is isolated, AI-Media’s LEXI Voice platform takes over–leveraging its proprietary AI workflow to generate highly accurate real-time captions and instant multilingual voice translations via synthetic AI announcers. The translated voices are then perfectly remixed over the original background audio, delivering a natural, lifelike broadcast in multiple languages that feels truly live and globally accessible.

The result is a fully automated, scalable solution that removes the cost and complexity of traditional multilingual production. This breakthrough enables rights holders and broadcasters to expand their reach, drive fan engagement, and monetize live sports like never before–all without sacrificing the energy of the original production.